The scholar draws on a mixture of original research—an experiment conducted among liberals and conservatives in Colorado—and secondary sources to navigate the civic space between political and social theory. He cites such behavioral-psych staples as Stanley Milgram’s authority/obedience tests, the Stanford prisoner experiment, and the dictator game. Yet Sunstein reassesses and sometimes challenges these landmark experiments. For example, he interprets Milgram’s key finding—the idea that most people crumple in the face of even mild authority—through the lens of group polarization, giving it a slightly rosier glow. Milgram’s subjects, Sunstein argues, were responding to assumed expertise on the part of the authority figure, rather than to authority itself. “Many of the subjects put their moral qualms to one side,” he writes, “not because of blind obedience, but because of a judgment that their qualms are likely to have been ill-founded.”

Yet very little, in the end, is completely new here. The specific concepts discussed in Going to Extremes—group polarization, informational cascades, rhetorical advantage—will be familiar to followers of Sunstein’s previous writings. Even the original research he cites (his Colorado experiment in particular) has been presented before. Nobody who has followed the author’s previous work on social theory and behavioral psychology will be surprised by his latest production.

But as is often the case among writers of Sunstein’s league, explanatory synthesis, and the elegance of a thesis proved, offer their own kind of value. What Going to Extremes lacks in terms of a broad, spectacular narrative, it makes up for in quiet comprehensiveness.

It helps that Sunstein’s prose is so lucid (which is not to be understood, in this case, as a polite euphemism for “mind-numbingly dull”). He writes with the measured cadence of a social scientist, in crisp sentences that go out of their way to explain themselves. And yet, underscoring Going to Extremes, and bubbling just below all the disciplined coherence, one senses an almost childlike fascination with the world and its workings, which both livens the book’s language and renders its arguments, if not more intellectually compelling, then at least more vividly charismatic than they would be were they articulated in the uniform vocabulary of social science. Occasionally an exclamation mark, like a rogue stowaway from Sunstein’s limbic brain, will make its way into an argument. (“Here is an especially disturbing finding. When people’s false beliefs are corrected, they might even become firmer in those beliefs!”) While some might call the scamp punctuation’s inclusion in the final manuscript a pander to the “pop,” I call it delightful.

Where Going to Extremes really shines, however, is in its other rare breakthrough moments: when it abandons explanation for exploration—when Sunstein uses his aggregated findings to take us somewhere new. And nowhere is this more true than in the book’s discussion of journalism. The tension that Sunstein establishes between the individual and the collective, between the peril and promise of group behavior, reflects an analogous strain in contemporary journalism: the informational desires of the individual versus the informational needs of a democratic citizenry.

The Web, for all the journalistic opportunities it offers, also enables the tyranny of the niche (Sunstein’s term is “cyberbalkanization”). The splintering of civic interest is quickly positioning itself as the bête-noire of digital-age democracy. It’s not only that the highly customized versions of journalism that are currently guiding the trajectory of innovation (RSS feeds! The Daily Me! mine magazine!) cede too much of the collective interest to the individual. They also deprive us of the simple serendipity of new information. And access to new information, Sunstein reminds us, is key—not merely to democracy, but to group dynamics more generally. “The most important reason for group polarization, and a key to extremism in all its forms, is new information,” he insists. “Group polarization often occurs because people are telling one another what they know, and what they know is skewed in a predictable direction. When they listen to each other, they move.”

Megan Garber is an assistant editor at the Nieman Journalism Lab at Harvard University. She was formerly a CJR staff writer.